I have data from several different frequency responses of the same open-loop system. This data has been put together to plot a Bode plot and Nyquist plot using MATLAB. I now need to find a transfer function that best represents the system. How would I go about deriving the transfer function? It can be done mathematically or using MATLAB, or trial and error though this is the less favourable approach.
I need to trial at least 1 first order and 1 second order transfer function using their standard forms to find which is the most suitable.
you can use the optimization techniques to find the best value of the transfer function
From `input-to-output' data, `system identification' is a technique to find system model. You may use System identification toolbox (MATLAB), which was useful for me. Under some constraint that you choose, you may get an optimized model.
https://mathworks.com/help/ident/ref/systemidentification-app.html
https://mathworks.com/help/ident/
Transfer function y(t)=x(t)*h(t).
basically from that we can determine the transfer function for signal x(t).
gain and phase margins are also calculated.
https://in.mathworks.com › answers
Using ARX to obtain a transfer function from frequency response data ...
I think Tree Po should get some basics in control theory and modeling in the frequency domain, and then a brief course in System Identification. If he knows what Optimization and ARX are from the beginning, he probably won't ask this question. There is no shortcut to achievement.
Tutorial on Introduction to System Identification by MATLAB
https://www.youtube.com/watch?v=u7hJ1aF-JrU
- Badges
- Science topic